Woulda loved to. Execs i traveled with had other plans tho.
That I totally get
. One of our top guys was in there to see how I did... Seriously though - I had people sitting on the FLOOR. That. was awesome.
![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
VMworld 2012 roll call...
- Thread starter RiDDLeRThC
- Start date
That I totally get
. One of our top guys was in there to see how I did... Seriously though - I had people sitting on the FLOOR. That. was awesome.
Thought your presentation was great!
Quick question..hopefully lopoetve can chime in here. For the HoL..I have the architecture document that VMware published..but i'm confused about one thing. I know they have a Control Center VM as the entry point to their vApp but they are using View so is the Control Center desktop within the vApp a View Desktop that's provided to the end-user or is the View desktop a completely different environment?
We are mapping out our Future Lab Infrastructure and while it certainly is not going to be as robust as this it will be based on vCloud Director and we would like to use some of the same methodology behind the HoL. We are going to use a customized version of Autolabs from labguides.com to provision a mobile lab environment to show our customers vSphere/View/VCD.
We are mapping out our Future Lab Infrastructure and while it certainly is not going to be as robust as this it will be based on vCloud Director and we would like to use some of the same methodology behind the HoL. We are going to use a customized version of Autolabs from labguides.com to provision a mobile lab environment to show our customers vSphere/View/VCD.
RiDDLeRThC
2[H]4U
- Joined
- Jun 13, 2002
- Messages
- 3,963
Am I the only one a little disappointed by the way they announced the new vCloud Suite in the GS.
Free Upgrade for Enterprise + customers, but they left out the little detail about that the free upgrade is to vCloud Suite Standard, and that vCloud Suite Standard doesn't include most of the things showcased in the GS.
Also with the announcement of vCloud Suite, what happens to vSphere Standard, Enterprise, Enterprise Plus? Is it sticking around? For us the cloud suite isn't really needed but we were looking at purchasing vCOPS.
Free Upgrade for Enterprise + customers, but they left out the little detail about that the free upgrade is to vCloud Suite Standard, and that vCloud Suite Standard doesn't include most of the things showcased in the GS.
Also with the announcement of vCloud Suite, what happens to vSphere Standard, Enterprise, Enterprise Plus? Is it sticking around? For us the cloud suite isn't really needed but we were looking at purchasing vCOPS.
Last edited:
RiDDLeRThC
2[H]4U
- Joined
- Jun 13, 2002
- Messages
- 3,963
Guess I will just need to run the numbers when I get back in the office. We were planning to upgrade to Enterprise Plus and also planned on buying vCenter Ops.
Some of the features in vCloud we are interested in so it may make sense to just upgrade to vCloud Suite Advanced.
Some of the features in vCloud we are interested in so it may make sense to just upgrade to vCloud Suite Advanced.
I keep hearing people say that vCloud Director is not for SMB's..yadda yadda yadda. I'll tell you that there are still cases for VCD in smaller customer orgs. We just implemented VCD to a smaller business where they provide security services to their customers, gov't..etc. Typically, they stand up environments and run testing with their custom applications..etc. With vCloud Director, they stand up the same environments same subnets/Ip's..etc all fenced and different developers can test different things against identical environments. Rapidly stand up their vApps and run them in a leased timeframe, and redeploy when needed..they love it.
There are use cases for vCloud Director for all sizes IMO.
There are use cases for vCloud Director for all sizes IMO.
Depends on your consolidation ratios, but the price for the Advanced suite is pretty reasonable since you're going to get vCOPS + vCD + Ent+
Agreed, it's a great lab manager replacement for dev environments... especially now that you can finally snapshot. I will admit that I haven't been exposed to it enough to see how it is used at the fullest potential.
side question: does snapshotting work with fast provisioned machines? I haven't seen anything that says otherwise... seems pretty facemelting to comprehend.
RiDDLeRThC
2[H]4U
- Joined
- Jun 13, 2002
- Messages
- 3,963
I'm going to bring my VAR in in the next few weeks to see if we can hash out a use case for our environment to help with justifying the price. I am interested in some of the vCloud Networking and Security pieces. If we can start replacing our firewalls then that would help recoup some of the costs.
Did you REALLY expect them to give you everything for the price of Enterprise+? Look at what's happening. The goal is to get everyone used to using vCD because that's the future. If you want the extras upgrade the bundles, which is cheaper than non-bundle pricing.
And regular vSphere and vCOPs licenses are still available.
Sorry, no clue - I wasn't involved in the insane (read: complex) HOL architecture
I know who to ask though. Let me find out.
RiDDLeRThC
2[H]4U
- Joined
- Jun 13, 2002
- Messages
- 3,963
For me I never really followed vCloud prior to the last week or so. I wasn't entirely sure what was included in the past and what wasn't.
Kinda got my hopes up earlier today when I stopped by the VMware booth and talked to someone about the licensing. He was under the impression that Enterprise customers weren't being left out and that they were included in the vCloud promo upgrades.
Kinda got my hopes up earlier today when I stopped by the VMware booth and talked to someone about the licensing. He was under the impression that Enterprise customers weren't being left out and that they were included in the vCloud promo upgrades.
RiDDLeRThC
2[H]4U
- Joined
- Jun 13, 2002
- Messages
- 3,963
Congrats, I sat in your first session and it was great. By far the best session I had.
Sorry, no clue - I wasn't involved in the insane (read: complex) HOL architecture
I know who to ask though. Let me find out.
If you can find out that would be awesome! Thanks.
Netjunkie,
Watched your session on the VMworld website, i'm sooooo glad they added the video content of some of the sessions already.
Crap, now I have 10 videos to watch as soon as I get home. Congrats on knocking it out of the park.
Do they sell access to the vworld site remote only? I could not make it out to vmworld this year but really wish I could catch some of these sessions online..
Scratch that.. It looks like all the videos are online free. Didn't used to be like this did it?
I just watched the Sakac and Stewart video. I'm curious why they suggest large VMFS datastores vs Duncan Epping that talks about using smaller one due to SCSI locking. When does KISS start affecting performance? At face value it seems like having a 64TB datastore is perfectly OK. What if that datastore has 200 VMs inside it? That seems counter-intuitive to everything else I read.
http://www.yellow-bricks.com/2011/07/29/vmfs-5-lun-sizing/
http://www.yellow-bricks.com/2011/07/29/vmfs-5-lun-sizing/
Last edited:
It depends.
With VAAI and Atomic Locking we can easily do a lot of VMs on large datastores. The reason I don't recommend huge datastores with 500 VMs on an array that can back it with good IOPS is...what happens if you lose a LUN for some mistake and need to restore? Takes a long time. Also, most array replication is LUN based and you probably don't want to have a single LUN of that size replicating.
I can understand that. Mitigating risk is a big one. Also replication I can see that as well. However wouldn't the replication concern only be an initial one? I guess it really depends how much change you have on your VMs, but shouldn't your replication technology only be migrating changes after the initial seed? I guess my point is if you have 10GB of change a day, wouldn't it still be 10GB of change regardless of LUN configuration if you replicate everything anyway?
I guess that breaking out into smaller LUNs allows you more granularity when it comes to 'what' you replicate and that's the bigger factor. With one large LUN, you're stuck migrating everything rather than just what really needs to be replicated.
As for restores, are you talking about recovery from backup or from a LUN-based snapshot because the snapshot should be fairly quick regardless of the size of the LUN. From a backup I can see that being an issue sure. I would think that anyone with the capability would be leveraging array based snapshots for faster recovery rather than needing to go back to a full backup right? Forgive me if my logic seems simple, I'm still getting my feet wet in the Storage arena.
I guess that breaking out into smaller LUNs allows you more granularity when it comes to 'what' you replicate and that's the bigger factor. With one large LUN, you're stuck migrating everything rather than just what really needs to be replicated.
As for restores, are you talking about recovery from backup or from a LUN-based snapshot because the snapshot should be fairly quick regardless of the size of the LUN. From a backup I can see that being an issue sure. I would think that anyone with the capability would be leveraging array based snapshots for faster recovery rather than needing to go back to a full backup right? Forgive me if my logic seems simple, I'm still getting my feet wet in the Storage arena.
For our smaller sites with low vm counts we create a datastore per VM and of course setup array based snapshots with EqualLogic. It works very well when they need to recover, had two customers recover quickly in the last month..they love it. Of course this is all based on the environment size..number of VM's..etc.
Exactly. Smaller LUNs let you replicate smaller data sets. Usually people don't replicate 100%. Also, often you set things like consistency groups in replication..so that LUNs stay in sync (or close to it). More overhead with bigger LUNs with more VMs that don't need to be in sync.
Snapshots depend on the original data. You blow away that LUN by accident or a failure and you now have no source data to rebuild that snapshot. Snapshots are for "oops, I deleted a file." not "oops, I deleted the wrong LUN.".
A 64TB datastore is a really bad idea, imho. That'll take days to restore when someone makes a mistake and blows it away.
Mitigate risk for RTO as well - don't forget how long it takes to rebuild BIG things vs small things.
And yes, it happens - I've gotten that call. Several times.
Even if you replicate off the snapshots, booting up 64T of VMs is a MUCH different proposition to booting up 500G worth and replicating them back. Assuming, like NetJunkie says, it didn't take the snaps/replication with it.
Can you? Sure. Should you? Personally, I don't think so.
Thanks guys for the input. These are really the same concepts that I was already working with prior to seeing that video. It complicates things for people with lesser experience when you hear things like that from veterans. It makes complete sense from a mitigation perspective to not do giant LUNs if you can avoid it while at the same time keeping the storage as easy to administer as possible. At the end of the day I control both the VMware side and the storage side of things, so making it harder only makes my job harder.
Watch my presentation - I cover a lot of the basics that get forgotten.
You got it, it's easier and quicker to revert the volume to a point in time snapshot than to mount and recover and considering it's one vm on the datastore you just roll back the volume. Very easy, very quick. It also accomplishes two things, one, you're not impacting any other VM that's on the datastore since it's dedicated to that VM in case of corruption or issues on the SAN side and of course as mentioned above, very easy, very fast to recover in the event of a OS FS corruption etc. The two customers that had issues the past month I was referring earlier was OS/Application related corruption, issues.
Is it couterintuative to what VMware stands for in the very nature of a pooled environment? I guess you could say that, it's not like this method is suggested in larger sites with hundreds of VM's. Typically, the largest we do this with is running maybe two or three host cluster, entry level EqualLogic like 4100 etc, and maybe 50 or less VM's.
Of course, I only recommend this in small environments. A lot of our business is with smaller mid-size and small business right now. This is not recommended for larger environments etc. I do agree with Netjunkie and lopoetve monster LUNS are NOT the way to go. Just because you can do 64TB Datastore doesn't mean you should.
DeChache
Supreme [H]ardness
- Joined
- Oct 30, 2005
- Messages
- 7,087
Do you know when yours will be posted? I can't seem to find it.
"soon".
Got great feedback on mine - lots of requests for some deep dives from the same "keep it simple" approach.
Last edited:
So I finally figured out how VMware access their HoL via View...you're presented with a View Desktop, however you're RDP'd from there to a Control Center VM within the particular Lab vApp.
After further research, it's not really the actual vApp's access,etc that is the hard part, it's the actual recreation of that thousands of times that's actually the magic. I won't have to worry about that really, as we'll most likely only have a couple of customers accessing the Demo Cloud at any given time.
VirtuallyGhetto actually explains how you can run nested ESXi within vCloud Director..as it's not listed as a valid OS within vCloud Director..so they'll have to be some table changes to the vCloud Director DB..not a big deal, Lam outlines it very well.
Within the vApp there will be a couple of nested ESXi hosts/vCenter/Domain-DNS Server/Virtual Storage Appliance, i'm guessing the Uber VSA since the relationship with EMC, a Control Center VM (Client that contains all the tools and utilities to access the lab), and whatever else is required for that lab itself. Then there are nested VM's within that..which is where more magic is happening. I'm going to try to script that out using vsphere powershell scripts that I can run when the vApp is up...this is all wrapped up and vShield, i'm guessing fenced to allow for same IP's...etc, and providing access to the vApp Control Center VM itself.
The automation framework is all custom..very neat..a lot of magic. but I won't have to really worry about that part. Also they are using Autodeploy..heavily so that's another thing that's going on I may dive into..not sure as of yet.
This is by no means a complete explanation, but it's the information that I needed to get out of it to leverage some of this for our Lab environment.
This is going to be a fun couple of weeks!
After further research, it's not really the actual vApp's access,etc that is the hard part, it's the actual recreation of that thousands of times that's actually the magic. I won't have to worry about that really, as we'll most likely only have a couple of customers accessing the Demo Cloud at any given time.
VirtuallyGhetto actually explains how you can run nested ESXi within vCloud Director..as it's not listed as a valid OS within vCloud Director..so they'll have to be some table changes to the vCloud Director DB..not a big deal, Lam outlines it very well.
Within the vApp there will be a couple of nested ESXi hosts/vCenter/Domain-DNS Server/Virtual Storage Appliance, i'm guessing the Uber VSA since the relationship with EMC, a Control Center VM (Client that contains all the tools and utilities to access the lab), and whatever else is required for that lab itself. Then there are nested VM's within that..which is where more magic is happening. I'm going to try to script that out using vsphere powershell scripts that I can run when the vApp is up...this is all wrapped up and vShield, i'm guessing fenced to allow for same IP's...etc, and providing access to the vApp Control Center VM itself.
The automation framework is all custom..very neat..a lot of magic. but I won't have to really worry about that part. Also they are using Autodeploy..heavily so that's another thing that's going on I may dive into..not sure as of yet.
This is by no means a complete explanation, but it's the information that I needed to get out of it to leverage some of this for our Lab environment.
This is going to be a fun couple of weeks!
Last edited:
That would sound like an amazing experience, especially if you can tie in a family vaca on the side.
What does the job think? Do they support you doing that?
"Go if you want to go!". I have approval to do it and it would be a good vacation..just very bad timing. I'd have to run down to Atlanta and get a passport quickly and shuffle some things around. And..my wife's father has been in the hospital the last week due to a stroke (he moved to the rehab center yesterday...not as bad as we feared). Sigh.